Bug 170185

Summary: sensors -s unreliable & BEEP problem
Product: [Fedora] Fedora Reporter: Nicolas Mailhot <nicolas.mailhot>
Component: lm_sensorsAssignee: Phil Knirsch <pknirsch>
Status: CLOSED DEFERRED QA Contact: David Lawrence <dkl>
Severity: medium Docs Contact:
Priority: medium    
Version: rawhideCC: rvokal
Target Milestone: ---   
Target Release: ---   
Hardware: All   
OS: Linux   
URL: http://www2.lm-sensors.nu/~lm78/readticket.cgi?ticket=2095
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2005-11-11 11:02:53 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:
Bug Depends On:    
Bug Blocks: 150221    

Description Nicolas Mailhot 2005-10-08 13:10:29 UTC
lm_sensors is very conservative with its fan divisor settings. It assumes low
resolution sensors which is less and less the case on modern hardware.

As a result the fan divisor value must be set with sensors -s to get a
meaningful fan reading. This operation seems unreliable - it works but not
always the first time, especially early in the boot process

sensors -s will not retry to set parameters in the background - you have to
script the retries
lm_sensors will no wait for the setting to succeed to monitor fan values. It
will compare the fan readings to the alert levels while they're still garbage
(because the divisor is still wrong)

As a result on my hardware linux boot = continuous LOUD alert beep. The
workaround is to do a 
/usr/bin/sensors -s
/usr/bin/sensors (sometimes you have to do a read for the setting to be taken
into account)

manually after the boot sequence which will set the divisor and shut up the
alert (put into the rc.local, only beeps a few minutes now)

This means you can't boot linux with other people around (office space, home box
late in the evening)

I had this problem with my previous mobo too (Via chipset, now nVidia one) -
except it was not beeping on alert, so the sensors reads were being silently wrong.

Please add a few retries and sleeps in the sensors init script. The current one
is fast but unreliable

Comment 1 Nicolas Mailhot 2005-10-27 18:32:54 UTC
See also: http://www2.lm-sensors.nu/~lm78/readticket.cgi?ticket=2095

Comment 2 Nicolas Mailhot 2005-11-11 11:02:53 UTC
After investigation upstream, the problem is some configuration directives in
sensors.conf were not in the order expected by sensors. And libsensors is too
dumb to reorder them by itself.

-> Long-term fix : enhance libsensors to automatically reorder configuration
directives or at least warn when they're not in canonical order